Thermal and structural analysis of Ni50Mn50−xInx shape memory alloys

In the present study, the Ni50Mn50−xInx (x = 12, 13 and 14 at.%) shape memory alloys were obtained by rapid solidification. The martensitic transformation and the solidification structures of these alloys were carried out by scanning electron microscopy, X-ray diffraction and differential scanning calorimetry, respectively. The experimental results showed that the crystalline structure of martensite in the In12 and In13 ribbons was identified as a 10M monoclinic structure, although the austenite has a cubic L21 structure for the In14 alloy. The martensitic transformation start temperature Ms decreases progressively with the increasing In content. The Ni content is mainly responsible for the adjustment in martensite transformation behavior in these shape memory alloys. Finally, the control of the valence electron by atom (e/a) determines the practical properties of these alloys at room temperature and makes it possible to create the alloys that can be candidates for various uses, such as sensors, refrigerants for magnetic refrigeration and actuators.


Introduction
Heusler-based Ni-Mn shape memory alloys (SMA s ) have received increasing interest in technological applications, due to their potential useful properties during martensitic transition (MT). This transition is from a cubic austenitic phase L2 1 or B2 at high temperature to a martensitic phase, whose structure can be L1 0 , 10M and 14M at low temperature [1][2][3][4]. Their properties make them of especially interest for the improvement in new magnetic sensors, actuators and refrigerants for magnetic refrigeration [5]. Experimental research proposes that the addition a third element to Ni-Mn alloys is a practical method to increase the martensitic transformation temperature [2,3]. The Ni-Mn-In system has potential significance as a shape memory alloy. The Heusler Ni 50 Mn 50-x In x alloys, with x within a particular scope of concentration values, have led to the observation of many interesting phenomena, for example, giant magnetocaloric effect (MCE) [6,7], large magnetoresistance (MR) [8,9], magnetic field-induced strain [10] and exchange bias [11]. Recently, scientists have considered the physical properties of Heusler Ni 50 Mn 50-x In x alloys. Krenke et al. [12] have detailed that the temperature M s decreases from 760 to 305 K, with the increasing In content. Besides, Sànchez-Llamazares et al. [13,14] demonstrated that the advancement temperatures of Heusler Ni-Mn-In alloys strongly depend on the composition, and their values spread in a very wide range. Furthermore, Sutou et al. [15] have announced that the martensitic transformation temperatures in the ferromagnetic state decrease gradually with the increasing In content. It is prominent that the compositional reliance of the Curie temperatures T c of the austenite phase is small, while the T c of the martensite phase strongly relies on the alloy compositions and decreases gradually with the decreasing In content in the Ni-Mn-In system. A [16][17][18][19]. The compositional reliance of the phase transition temperature is also credited to the adjustment in electron concentration and the Mn-Mn interatomic distance [20,21]. Presently, most SMA s are Ni-Mn based and can be obtained by the substitution of some main group elements (Ga) for Mn or Ni in Ni-Mn alloys [22]. This substitution brings down the martensitic transformation temperature of Ni-Mn effectively and results in a highly ordered Heusler-type structure. This makes it possible to predict the martensite transformation by first-principles calculations in Ni-Mn-based Heusler alloys.
In this paper, we carried out an experimental study on microstructure and phase transformation behavior of Ni 50-Mn 50-x In x (x = 12, 13 and 14 at.%) alloys. The change in In content was found to affect the martensitic transition temperatures of these shape memory alloys.

Materials and methods
As-cast ingots of about 2 g with nominal composition Ni 50 Mn 50-x In x (x = 12, 13 and 14 at.%) were prepared by conventional argon arc melting from Ni metal filament (purity [ 99.98%), a metallic manganese sheets (purity [ 99.98%) and pieces of metallic In of small spheres (purity [ 99.99%), using Bühler MAM-1 compact arc melter. These ingots were melted four times to ensure a good starting homogeneity. The samples were induction melted in quartz crucibles with a circular nozzle of 0.6 mm and ejected by applying an argon overpressure on the polished surface of a rotating copper wheel * 48 ms -1 . The as-spun ribbons were flakes of 1.2-2.0 mm in width and several centimeters in length. The prepared samples are named as follows: In12, In13 and In14, respectively. The microstructure of the alloys was examined by using scanning electron microscopy (DSM960A; Zeiss, Norman, OK, USA; SEM) equipped with energy-dispersive X-ray spectrometry. X-ray diffraction (Bruker D8 Advance diffractometer in a 2h-geometry; Manning Park Billerica, MA, USA; XRD) analyses were performed using Cu-K a radiation. The structure of ribbons was refined by the Rietveld method using Maud Program and Jana software (Jana 2006, Jana, Praha, Czech Republic) [23,24]. The martensitic transformation temperatures were measured by means of differential scanning calorimetry (DSC822 apparatus of Mettler Toledo; Columbus, OH, USA; DSC) instrument with a heating/cooling rate of 10 K min -1 under constant argon flow.
Results and discussion Figure 1 shows the micrographs corresponding to the free, wheel surface sides and typical fracture cross section at various magnifications of the ribbons. The typical SEM images of the wheel surfaces of In12, In13 and In14 alloys are shown in Fig. 1a-c, respectively. The wheel surfaces of the In12 and In13 alloys clearly present the lamellar microstructure of the martensite structure (Fig. 1a, b). These ribbons are mechanically fragile and brittle and are effective along the perpendicular direction to the ribbon plane, while the wheel surface of the In14 alloy is represented by a granular microstructure of the austenite structure ( Fig. 1c). At the free surface in Fig. 1d-f, the samples are given a microstructure consisting of granular, columnar grains growing through the ribbon thickness with their longer axis aligned perpendicular to the ribbon plane. The grain size varied somewhere in the range of 1 and 3 lm. This result has been previously observed in Ni 45.8 Mn 42.6-In 11.6 alloy ribbons [25] and confirmed by González et al. [26]. A microcrystalline microstructure is observed in the inside of each grain for the presence of various zones where the solidification of the material ends. The thickness of each ribbon is determined from the average values measured on several micrographs (see Fig. 1i-k). It can be seen that the values oscillate between 10 and 17 lm. A rough comparison of their cross-sectional columnar-like microstructure suggests that the presently produced Ni-Mn-In alloys exhibit quicker grain development energy, as revealed by the larger in-plane grain width that roughly varies between 1 and 8 lm [13].
The obtained results of EDX microanalysis of the chemical composition examined by scanning electron microscope (SEM) show a similarity with the hypothesis of the chemical composition of the studied as-spun ribbons ( Fig. 2a-c).
It was observed that the composition analysis was in perfect agreement with the nominal compositions of the as- To determine the conditions of thermal analysis, the knowledge of crystal structure at room temperature is basic [27,28]. The XRD diagrams of the In12, In13 and In14 melt-spun ribbons at room temperature are shown in Fig. 3. The XRD patterns of In12 and In13 ribbons illustrate a martensite phase of 10M monoclinic structure with lattice parameters: a = 4.31 ± 0.003 Å , b = 5.68 ± 0.006 Å , c = 21.002 ± 0.009 Å with an angle b = 87.53°for In12 alloy and a = 4.34 ± 0.004 Å , b = 5.72 ± 0.005 Å , c = 21.019 ± 0.008 Å with an angle b = 87.58°for In13 alloy. These parameters are in good concurrence with those recently detailed for melt-spun ribbons of close chemical composition [2]. Both diffraction patterns were indexed and identified to the monoclinic structure using Rietveld analysis via the Jana software. Miller indexes were assigned with the aid of indexing program as Treor and Dicvol. The structure of the In12 and In13 alloys was almost identical, with the exception of little differences in the peak intensities and angles. These results are consistent with electron microscopy observations of the martensitic structure, whereas the In14 alloy has a L2 1 cubic structure with lattice parameters: a = b = c = 5.987 ± 0.004 Å . The structure L2 1 is confirmed by the observation of some superstructure peaks, such as reflection peaks (3 1 1) and (5 3 1) [2]. All the results in these three samples are steady with those Ni-rich Ni-Mn-In alloys acquired by other authors [1]. When decreasing the In content, the stable phase at room temperature turns into an austenitic phase. Thus, these structural properties of both the austenite and the martensite are found to change with the increasing In content, clearly identified with the smaller size of Mn atoms (r a = 0.140 nm) relative to that of In atoms (r a = 0.167 nm). It appears that the considerably larger size of the indium atoms as compared to other group III-A and group IV-A elements leads to an Mn-Mn separation that is quite large [12]. In addition, with the increasing In content, the excess Mn atoms occupy In sites. In such spatial configurations, Mn-Mn neighbors have a smaller separation than that in the stoichiometric compound [12]. In this way, we can conclude that the variation in the crystal structure at room temperature from the structures 10M to austenitic cubic L2 1 depends on the In composition. Another factor, probably, as the adjustment of either the production conditions or the little changes in composition supports the thermal stability of various structures and, as a result, different magneto-elastic behaviors. The small and constrained grains in the flakes Thermal and structural analysis of Ni 50 Mn 50-x In x shape memory alloys 3067 might have made the transition to the martensite phase difficult and moved it to lower temperatures, most likely connected with the increased degree of quenched-in shortrange disorder around defects, as proposed by Chernenko et al. [29]. Obviously, factors such as the slight move in the valence electron concentrations additionally increase the structural complexity. Based on the XRD results, it is clear that the DSC scans of the In12 and In13 alloys should be performed by heating at room temperature in order to detect the martensitic transition. Likewise, the DSC scan of the In14 alloy could be performed by cooling from room temperature. The relating DSC results are given in Fig. 4.
The martensitic transformation temperatures were determined by the crossing point of the base line and the tangent line which is the largest slope of the peak, as represented in Fig. 4. The obtained transformation temperatures are recorded in Table 1.
The hysteresis DT (DT = A s -M f ) is due to the increase in the elastic and the surface energy during the martensite formation. In this manner, the nucleation of the martensite suggests super-cooling. Along these lines, it chose the width of the hysteresis, DT, as the difference in the  temperatures relating to the peak position, 0.5, 13 and 14 K for alloys In12, In13 and In14, respectively. The transformation region can likewise be described by the martensite transformation temperature T 0 (T 0 = 9 (M s ? A f )): the temperature at which the Gibbs energies of the martensite and parent phases are equal [30]. This parameter of three samples is collected in Table 1.
The entropy (DS) and enthalpy (DH) changes in the structural transformations are calculated from calorimetry data [12] using the relationships: and where T i and T f are the temperature limits of integration. The enthalpy and entropy changes' values are likewise incorporated in Table 1. We can conclude that, for both DS and DH, there is no significant concentration dependence. Nonetheless, these thermodynamic parameters take their highest values for an In 13 alloy, while, for the In12 and In14 ribbons, they have the least values of DS and DH. Krenke et al. [12] detailed that these thermodynamic parameters DS and DH decrease with the increasing In concentration. They related this decrease to the large difference in the magnetic exchange interactions below and above M s , which gave rise to a positive magnetic entropy change in the proximity of martensitic transformation. The reason of the differences has found between samples with different values of the injection over pressure or the distance between wheel and injection point [31]. On the other hand, one parameter used to characterize shape memory alloys is the valence electron concentration (e/a) that is calculated using the electron concentration of the outer shells for each element of the Ni-Mn-In system. The (e/ a) ratio is determined as: (e/a) = [10 9 (Ni at.%) ? 7 9 (Mn at.%) ? 3 9 (In at.%)]/100, where 10, 7 and 3 represent the number of valence electrons for Ni (3d 8 4s 2 ), Mn (3d 5 4s 2 ) and In (5s 2 4p 1 ), respectively [12].
We note that the values decrease as the indium content increases (decreasing (e/a)). A similar pattern was found by other authors. The crystal structure of martensite formed in Ni-Mn-In alloys varies with alloy composition (or (e/a)), and the change in the unit cell volume caused by the structural transformation would be distinctive [32], therefore leading to the change of DS. In addition, it is known that there is a linear correlation between the average number of valence electrons per atom and the martensite start temperature M s ; it increases/decreases when the value (e/a) increases/decreases [12,[27][28][29]. Similar behavior is found in this study, where the M s decreases from 431 to 260 K when (e/a) shifts from 8.02 to 7.94 for In12 and In14, respectively. Thus, the control (e/a) decides the transformation temperatures range, in this type of alloys, and it is possible to develop alloys with the desired transformation temperatures as candidates for some applications such as sensors and actuators.
In addition to the chemical composition, the atomic order of the parent phase also has a great influence on the martensitic transformation. In this way, a significant reduction in the transformation temperature extension, possibly related with nearby differences in atomic order, is obtained. For example, Kreissl et al. [33] found that the disorder existing between Mn and Ga atoms in Ni 2 MnGa alloy considerably decreased M s by about 100 K. On the other hand, what is interesting in our study, we can see that the peaks of In12 and In13 alloys are progressively exceptional and exhibit a remarkable expansion and amplitude compared to those of In14 alloy. In this manner, we also noted in Table 1 that both the enthalpy and entropy changes decreased due to the increasing In content. This outcome is similar from that reported in Ref. [12] in which the enthalpy and entropy change related to the transformation decreases as (e/a) ratio decreases. The discrepancy between this study and Ref. [12] remains unclear, but maybe based on the literature, the impact of (e/a) ratio on the enthalpy and entropy change of martensitic transformation has announced for Ni 2?x Mn 1-x Ga alloys [34,35] Thermal and structural analysis of Ni 50 Mn 50-x In x shape memory alloys 3069 a) reliance of DS is related to the magnetic contribution that depends on the difference in the magnetic exchange below and above M s [34]. It is generally known that the magnetic exchange is closely associated with the Mn-Mn interatomic distance. In any case, it has been appeared that there is no significant concentration dependence for other systems, for example Heusler Mn-Ni-Sn, Cu-Al-Mn and Ni-Mn-Sn-Co alloys [36][37][38]. On the other hand, as shown in Fig. 4, with an increase in the indium content, the DSC scan was performed near room temperature and the cyclic process was additionally close to room temperature. As expected, the austenite state for these alloys ends at room temperature because the martensite state starts at this temperature, which permits the resolution of the structure of martensite product [39]. Likewise, from a thermodynamic point of view, there is irreversible entropy, DS i , at the hysteretic transition. This irreversible entropy is relative to the thermal hysteresis. It can be determined by DSC data taking into account the peak temperature of the austenite to martensite, T AM , and the martensite to austenite, T MA . The equation as given in Ref. [40] is: The values (DS i /DS) are less than -0.02. Therefore, there is a small correction, because the differences obtained in DS by applying various ways are higher (see Table 2). Similar behavior has been found in the literature in the Ni-Co-Mn-Sn system [40]. Furthermore, in the transformation energy, there is a term related to the effect of the elastic contribution, DE el , which can be calculated by applying the equation in Ref [31].
It is noted that the values (DE el /DH) are low, except for the sample In14. This alloy has also lower values of enthalpy and entropy. It is also known that the presence of different martensitic variants tends to minimize the elastic energy associated with the deformation of the sample. Thus, lower martensitic variants require higher energy expenditure to initiate the martensitic transformation [41].
All these results are related to a decrease in martensite variations in the samples.

Conclusions
The microstructure and martensitic transformation behavior of Ni 50 Mn 50-x In x (x = 12, 13 and 14 at.%) alloys were explored by SEM, XRD and DSC. The crystal structure of martensite in the ribbons is identified as a 10M monoclinic structure, although the austenite has a L2 1 cubic structure. The microstructure has noted that the grain sizes increase with the increasing In content. The micrographs showed the presence of equi-axial and columnar grains with an inhomogeneous distribution, whereas the chemical composition was homogeneous. The increase in the grain size is accompanied by a decrease in the M s, demonstrating that the structural transition temperatures can be adjusted within certain limits by controlling this microstructural parameter. We note that this M s increases for nominal (e/ a) decreases with the partial substitution of Ni by In. The (e/a) control makes it possible to improve the alloys at the ideal transformation temperatures. Likewise, the entropy and enthalpy change related to the transformation decreases as (e/a) decreases. Finally, the control of the valence electron by atom (e/a) determines the practical properties of these alloys at room temperature and allows to create alloys that can be candidates in a various uses, such as sensors and actuators.